Create random smiles

randomsmiles.py

import random
import pybel

def create_mutants(A, B):
    # Let's randomly choose a cross-over point in both A and B
    # and generate four possible combinations
    c1 = random.randint(0, len(A))
    c2 = random.randint(0, len(B))
    startA, endA = A[:c1], A[c1:]
    startB, endB = B[:c2], B[c2:]
    children = [
            startA+endB, startB+endA, # somewhat sensible
            endA+startB, endB+startA, # less sensible
            ]
    # Let's mutate a few characters by swapping nbors randomly
    mutant_children = []
    for child in children:
        mutant = ""
        i = 0
        N = len(child)
        while i < N:
            if i+1 < N and random.random() > 0.66: # 1 in 3 chance
                mutant += child[i+1]
                mutant += child[i]
                i += 1 # extra increment
            else:
                mutant += child[i]
            i += 1
        mutant_children.append(mutant)
    random.shuffle(mutant_children) # don't favour any of them
    return mutant_children

def get_mutant(smiA, smiB):
    for N in range(50): # try combining these 50 times
        mutantsmis = create_mutants(smiA, smiB)
        for mutantsmi in mutantsmis:
            try:
                mol = pybel.readstring("smi", mutantsmi)
            except IOError:
                continue # bad syntax
            return mutantsmi, mol
    return "", None

if __name__ == "__main__":
    with open(r"C:\Tools\LargeData\chembl_23.smi") as inp:
        allchembl = inp.readlines()
    dataset = [x.split("\t")[0] for x in random.sample(allchembl, 1000)]

    maxwt = 0
    while True:
        smiA, smiB = random.sample(dataset, 2)
        mutantsmi, mol = get_mutant(smiA, smiB)
        if mol and mol.molwt > maxwt:
            print(mutantsmi)
            print(mol.molwt)
            maxwt = mol.molwt
            print()